The use of local oscillator generation circuits for wireless transceivers is well known in the art. The local oscillator is generated as a continuous wave (CW) and is then used for quadrature modulation or demodulation of transmitted and received signals respectively. Alternatively, the oscillator can also perform frequency modulation as part of a polar transmitter architecture system.
A block diagram illustrating an example prior art phase locked look (PLL)-based local oscillator (LO) generator circuit is shown in FIG. 1. The typical PLL LO generation circuit, generally referenced 10, comprises phase detector (PD) 14, loop filter or low pass filter (LPF) 18, controlled oscillator 22, resonator 26 and frequency divider 28.
In operation, a reference signal 12, normally generated by a crystal oscillator, is input to the phase detector 14 along with a divided-down RF frequency continuous wave (CW) 29. The phase detector, typically implemented as a charge pump or mixer, generates a phase error (PE or PHE) 16 proportional to the phase difference between the fREF input signal 12 and RF CW signal 29. The resultant PE signal is then low pass filtered using low pass filter 18 to yield a slow varying frequency command signal 20.
The frequency command signal is input to a controlled oscillator circuit 22, typically a voltage controlled oscillator (VCO) or a digitally controlled oscillator (DCO). This oscillator generates an RF signal 24, the frequency of which roughly depends linearly on the frequency command signal. The oscillator uses a resonator 26 that oscillates in the desired frequency band. Resonator circuits can be inductor-capacitor based (LC) or closed loop inverter chains (ring). The output of the oscillator 22 is the phase locked LO signal fLO or fRF which also undergoes division by N using divider 28 to generate the feedback signal 29 to the phase detector.
A major problem associated with LO generation schemes such as that of FIG. 1 is their susceptibility to RF signal interference. In particular, the resonator used in the circuit (especially inductor based resonators) often picks-up unwanted RF signals and the resonator frequency can be severely perturbed. This phenomenon is known as frequency pulling and is defined as an effect that forces the frequency of an oscillator or resonant frequency to change from a desired value. Causes of the pulling include undesired coupling to another frequency source (e.g., RF intermediate or output signals) or the influence of changes in the oscillator load impedance. Typically, the interferer is either the modulated amplified output RF signal, its harmonics in transmitters or the amplified received signal in receivers. To avoid frequency pulling, a well defined RF transceiver system is built such that the actual resonation frequency of the resonator is neither the output RF frequency, nor any of its harmonics or sub-harmonics.
In the case of a mobile wireless system, for example, transmitters that modulate a non-constant envelope signal require a non-integer ratio between the local oscillator frequency and the RF frequency in order to overcome the pulling effect of the power amplifier's output harmonics. Transmission of a wideband signal in high frequency bands such as 5 GHz, however, requires complicated converters that run at very high frequencies.
A block diagram illustrating an example prior art ½X local oscillator generation scheme is shown in FIG. 2. The example circuit, generally referenced 170, comprises a synthesizer at ½fRF, X2 frequency doubler 176 and polyphase filter 180. In this example LO generation circuit, the input reference frequency fREF 171 is input to synthesizer 172 tuned to exactly ½ the RF frequency ½fRF. The output signal 174 is then input to a frequency doubler 176 to generate a signal at fRF. This signal is then filtered via polyphase filter 180 to yield I and Q (i.e. separated by 90 degrees, also referred to as quadrature) output clock signals fLOI 182 and fLOQ 184, respectively, at fRF. The polyphase filter is needed in order to generate the quadrature output signals. An advantage of this scheme is the fact that the actual oscillation frequency is not the final output frequency but is half. Although the circuit generates fRF signals, a major disadvantage of using the polyphase filter is that they are typically large and inaccurate filters causing a potentially large IQ mismatch, i.e. LOI and LOQ are not strictly 90 degrees apart. If such a synthesizer solution is inductor based then halving the frequency forces the size of the inductors to increase significantly.
A block diagram illustrating an example prior art 2X local oscillator generation scheme is shown in FIG. 3. The well known and widely used LO generation scheme (2X scheme), generally referenced 190, comprises synthesizer 194 and frequency divider 198. A crystal oscillator generated reference signal 192 is input to a synthesizer 194 tuned exactly to twice the RF frequency (2fRF). The resultant output signal 196 is then divided by two using a frequency divider 198 to generate two signals having a quadrature relationship, i.e. I and Q output signals fLOI 200 and fLOQ 202, respectively, at fRF.
These signals can be used to modulate or demodulate a signal using a mixer pair in a zero IF (ZIF) or a near zero IF (NZIF) scheme. The advantages of this scheme is the fact that the actual oscillation frequency is not the final output frequency but its double and that it is relatively easy to generate a clean quadrature pair fLOI and fLOQ using a frequency divider 198.
Two major disadvantages of this scheme, however, are (1) the fact that the second harmonic of the amplified RF signal at 2fRF can pull the oscillator away, since there could be a small offset between these two frequencies due to data modulation and (2) that the oscillator must be designed to twice the frequency (generally design at high frequencies tends to be more difficult). The first disadvantage can manifest itself in second harmonic leakage from the system output coupling back into the heart of the resonator or the first harmonic coupling back into the synthesizer supply circuitry and generating the second harmonic using a non-linear effect and creating frequency pulling. Another manifestation of this disadvantage can be in the receiver where a high gain version of the input signal at fRF, when compressing a certain stage of the reception chain can create a second harmonic, which will also pull the oscillator (i.e. injection pulling or, worse, injection locking). Injection locking occurs when the oscillations of a first system influences a second system to the extent where the second system no longer oscillates at its own natural frequency but rather at the frequency of the first system. In the case of injection pulling, the second system can still oscillate at its own natural frequency, but contains energy at the frequency of the first system. For near-zero IF systems, such injection locking can cause the oscillator to be pulled down or up to the actual RF frequency thus making the system effectively a poorly designed zero-IF system.
To avoid these disadvantages, the LO can be generated at a rational multiplier of the output RF frequency. A block diagram illustrating an example prior art local oscillator generation scheme that generates the LO at a rational multiplier ( 4/3fRF in this example) of the output RF frequency is shown in FIG. 4. The prior art LO generation circuit, generally referenced 210, generates the LO at a rational multiplier of the output RF frequency and uses dividers and mixers to generate the output RF frequency. The circuit 210 comprises a synthesizer 214, frequency dividers 216, 220, multipliers 222, 224 and band pass filters (BPF) 226, 228.
The scheme of FIG. 4 is typically known as an offset-LO generator. A crystal oscillator output reference signal 212 is input to a synthesizer (PLL) tuned to exactly 4/3fRF. Its output signal is divided by two using frequency divider 216 to yield a signal at ⅔fRF 218. This signal is divided by two again using frequency divider 220 to yield a quadrature signal pair 221, 223 at ⅓fRF. Signals 221, 223 are mixed with signal 218 separately via analog mixers 222, 224, respectively. Due to the multiplicative nature of the mixer it generates a product at fRF (its inputs having frequencies of ⅓fRF and ⅔fRF) while signals 221, 223 also have a 90 degree phase difference at fRF and thus constitute a quadrature pair. Since the mixer is not ideal, however, undesired frequency products at n/3fRF (where n is an integer, n≠3) will also be present at the output of the mixers. Band pass filters 226, 228 attenuate these unwanted products yielding the final LOI (fLOI) 230, LOQ (fLOQ) 233 signals, respectively.
An advantage of the offset LO scheme 210 is that it is able to generate an LO signal at fRF, while the resonator oscillates at a rational multiple of fRF rather than an integer multiple. Hence, no harmonics of the output frequency can interfere with the proper operation of the oscillator. While this circuit generally avoids the frequency pulling phenomena described supra, it has a significant disadvantage in the unwanted products (i.e. spurs) generated by the mixers. These products likely cause spectral emission mask (SEM) violations in the transmitter and can downconvert unwanted jammers or blockers in the receiver. Hence, the spur attenuation or filtering requirement for BPFs 226 and 228 is usually very significant.
It is thus desirable to have a local oscillator generation mechanism that overcomes the disadvantage of the prior art techniques. The local oscillator generation mechanism should preferably be implementable as an all digital circuit and oscillate at a rational RF frequency multiplier (n/m fRF) so as to avoid frequency pulling while reducing or alleviating the need for a stringent BPF. Further, the local oscillator generation mechanism should enable wideband modulation, such as for polar modulation, requiring a relatively simple, all digital implementation.